A voting model weighting algorithm-driven multi-algorithm framework enhances intelligent point-of-care testing for edible oil safety

Edible oils, a major source of dietary fat, are frequently contaminated with dibutyl phthalate (DBP) and benzo( a )pyrene (BaP), posing serious food safety risks. To enhance the accuracy of immunoassays in complex oil matrices, we developed a point-of-care testing platform that integrates bimetallic porous carbon nanomaterials as multifunctional signaling probes with a multi-algorithm framework driven by a voting model weighting algorithm. The porous carbon material Zn-CN was synthesized via organic solution coordination followed by high-temperature pyrolytic derivatization. Pt-Zn-CN was subsequently prepared by incorporating platinum nanoparticles into Zn-CN through reductive stirring at room temperature. This nanomaterial exhibits excellent photothermal properties and dispersibility, and its large surface area and porous architecture facilitate efficient antibody coupling and target enrichment. The resulting nanoprobes act as multifunctional signal transducers to enhance the recognition and amplification of antigen–antibody interactions. The multi-algorithm framework leverages bionic swarm intelligence and autonomous decision-making to perform high-precision image segmentation, attaining 97.6 % contour recognition and 95.7 % detection accuracy within 0.03 s. This platform enables ultra-fast early warning and achieves ultra-low detection limits of 0.184 ng/mL for DBP and 0.096 ng/mL for BaP. A user-friendly graphical user interface supports real-time quantification of BaP and DBP, providing intuitive safety warnings for edible oils.